2.1 The H19 Gene
The H19 gene is one of the few genes known to be imprinted in humans (Hurst et al., 1996, Nature Genetics 12:234-23). At the very beginning of embryogenesis, H19 is expressed from both chromosomal alleles (DeGroot et al., 1994, Trophoblast 8:285-302). Shortly afterwards, silencing of the paternal allele occurs, and only the maternally inherited allele is transcribed.
H19 is abundantly expressed during embryogenesis, and was first identified as a gene that was coordinately regulated with alpha-fetoprotein in liver by the trans-acting locus raf (Pachnis et al., 1984, Proc. Natl. Acad. Sci. USA 81:5523-5527). Additionally, H19 has been independently cloned by a number of groups using screens aimed at isolating genes expressed during tissue differentiation. For example, Davis et al. (1987, Cell 51:987-1000) identified the mouse homolog of H19 in a screen for genes active early during differentiation of C3H10T1/2 cells. Pourier et al. (1991, Development 113:1105-1114) found that murine H19 was expressed during stem cell differentiation and at the time of implantation. Transcription of the human H19 gene was also discovered in differentiating cytotrophoblasts from human placenta (Rachmilewitz et al., 1992, Molec. Reprod. Dev. 32:196-202).
While transcription of H19 RNA occurs in a number of different embryonic tissues throughout fetal life and placental development, H19 expression is down-regulated postnatally. Relatively low levels of H19 transcription have been reported, however, in murine adult muscle and liver (Brunkow and Tilghman, 1991, Genes & Dev. 5:1092-1101). H19 also is activated postnatally in cancer cells. Ariel et al. (1997, Molec. Pathol. 50:34-44) demonstrated H19 expression in a number of tumors arising from the tissues in which it is expressed prenatally. Additionally, these authors found H19 RNA in tumors derived from neural tissues, in particular astrocytoma and ganglioneuroblastoma, that are not known to be associated with H19 expression. Given the large array of cancers expressing H19 RNA, these authors speculated that H19 was an oncofetal RNA and proposed investigating H19 as a tumor marker for human neoplasia.
Both human and murine H19 genes have been cloned and sequenced (Brannan et al., 1990, Molec. Cell. Biol. 10:283-6). Comparison of the human and mouse H19 genes revealed an overall 77% nucleotide sequence identity. Despite this conservation of nucleotide homology between species, very low deduced amino acid sequence identity could be predicted from the open reading frames of the two genes (Id.). Further, although H19 RNA is transcribed by RNA polymerase II, spliced and polyadenylated, it does not appear to be translated. Instead, H19 RNA has been found associated with the 28S cytoplasmic RNA, leading to speculation that H19 RNA may function as an RNA component of a ribonucleoprotein (Id.).
The actual physiological role of H19 is not fully understood. H19 can act as a dominant lethal gene; a high ectopic expression of an H19 transgene causes lethality in mice shortly before birth (Brunkow et al., supra). This lethal period coincides with the time when H19 transcription becomes repressed. On the other hand, no defect has been observed in either heterozygous or homozygous mice carrying an H19 knockout allele(s) (Leighton et al., 1995, Nature 375:34-39). A knockout of the maternally inherited allele does interfere with imprinting of the genetically linked and oppositely imprinted IGF-2 gene; the resulting mice are larger at birth than their littermates due to the increased prenatal expression of IGF-2 (Id.). Since these two oppositely imprinted genes share cis-acting regulatory sequences, Leighton and colleagues speculated that H19 may be involved in imprinting the IGF-2 gene.
Another function proposed for the H19 gene product is that of a tumor suppressor RNA. Hao et al. (1993, Nature 365:764-767) reported that transfection of two embryonal tumor cell lines, RD and G401, with an H19 expression construct resulted in cell growth retardation, morphological changes and reduced tumorigenicity in nude mice. Such a tumor suppressor activity has been noted as consistent with the observed lethality of ectopic expression in mice (Hao et al., supra) as well as the increased size of mice that are knocked out for the maternal H19 allele (Leighton et al., supra). The proposal that H19 is a tumor suppressor has been controversial, however. Some of the results were reportedly not reproduced, and there may exist another candidate tumor suppressor gene closely linked to H19 (Ariel et al., supra). H19's proposed role as a tumor suppressor also conflicts with the experimental data that H19 is activated in a broad array of tumor cells (see for example Lustig-Yariv et al., 1997, Oncogene 23:169-177).
2.2 The Insulin-Like Growth Factor (IGF) Genes
IGF-2 is another imprinted gene whose expression depends upon it's parental origin. However in contrast to H19, IGF-2 in both mice and humans is maternally imprinted and therefore expressed from the paternally inherited allele (Rainier et al., 1993, Nature 363:747-749). The human IGF-2 gene exhibits a complex transcriptional pattern. There are four IGF-2 promoters that are activated in a tissue and developmentally specific manner. Only three of the promoters, P2, P3 and P4 are imprinted and active during fetal development and in cancer tissues. The fourth, promoter P1,is not imprinted and is activated only in adult liver and choroid plexus (see Holthuizen et al., 1993, Mol. Reprod. Dev. 35:391-393). The P3 promoter of the IGF-2 gene has been implicated in the progression of liver cirrhosis and hepatocellular carcinoma (Kim and Park, 1998, J. Korean Med. Sci. 13:171-178).
Loss of imprinting of IGF-2 has been implicated in Wilm's tumor (Ogawa et al., 1993, Nature 363:749-751). This observation led many investigators to speculate that the loss of imprinting and biallelic expression of imprinted genes may be involved in growth disorders and the development of cancer (see also Rainier et al., 1993, Nature 362:747-749, and Glassman et al., 1996, Cancer Genet. Cytogenet. 89:69-73).
2.3 Tumor-Specific Gene Therapy
Regulatory sequences from tumor-associated genes have been used to selectively target expression of a suicide gene in tumor-derived cells. For example, alpha-fetoprotein expression is induced in hepatocellular carcinoma. Huber et al. (1991, Proc. Natl. Acad. Sc. USA 88:8039-8043) used control sequences from either the albumin gene or the alpha-fetoprotein gene to target expression of varicella-zoster thymidine kinase (VZV TK) gene coding sequences to hepatoma cells. Hepatoma cells infected in vitro with a retroviral vector containing one of these expression constructs expressed VZV TK and became sensitive to the normally non-toxic prodrug 6-methoxypurine arabinonucleoside (araM). Kaneko et al. (1995, Cancer Res. 55:5283-5287) constructed an adenoviral vector expressing HSV TK under the control of the alpha-fetoprotein control sequences. Recombinant adenoviral particles containing this vector were directly injected into hepatocellular carcinoma-derived tumors generated in athymic nude mice. Subsequent intraperitoneal injections with ganciclovir caused regression of the hepatocellular carcinoma-derived tumors.
Osaki et al. (1994, Cancer Res. 54:5258-5261) transfected into A549 lung carcinoma cells an expression construct containing the control sequences for the lung carcinoembryonic antigen gene linked to the coding sequence for Herpes simplex virus thymidine kinase (HSV TK). The transfected cells were sensitive to ganciclovir. Additionally, tumor growth in nude mice from subcutaneously injected transfected cells was inhibited by repeated intraperitoneal injections of ganciclovir. However, the carcinoembryonic antigen gene has recently been described as expressed in normal colonic mucosa, thus limiting the usefulness of these control sequences as tumor specific regulatory regions (Osaka et al., supra). Thus, there remains a need for the development of gene therapy vectors that specifically express gene products in tumor cells.